Use best practice configurations for the SharePoint 2013 virtual machines and Hyper-V environment
Applies to: SharePoint Server 2013, SharePoint Foundation 2013
Topic Last Modified: 2014-03-10
Summary: Follow best practice recommendations to configure the SharePoint 2013 virtual machines and the Windows Server 2008 Hyper-V infrastructure.
After you create a detailed architecture design and system specifications, you are ready to install and configure the virtual environment for the SharePoint 2013 farm. To achieve the performance and capacity goals that you identify in your detailed design and system specifications, you must have a correctly configured virtual environment. One or more badly configured virtual machines or virtualization hosts can significantly reduce performance.
This article discusses Windows Server 2008 Hyper-V technology configuration options and their potential impact on the performance of the virtualization host computer, the virtual machines, and the SharePoint 2013 farm in general. Key decision points and best practice configurations provide guidance when you set up the virtual environment.
|A wide range of features and installation options are available to set up and configure a Hyper-V environment. Measurements focus on testing and guidance on the features and configurations that provide quantifiable benefits and consider the known performance characteristics of server roles in a SharePoint products farm.|
In this article:
We recommend that you use a bottom up approach to configure the Hyper-V environment for the SharePoint 2013 farm. Start with the Hyper-V host computer configuration and then work up to the virtual layer to configure the virtual network components and the virtual machines.
Before you install and configure the Hyper-V environment, we recommend that you review the following articles:
Several aspects about how to configure a virtual environment that are not discussed in detail are documented in separate articles. Detailed information about the following subjects is out of scope for this article:
Business Continuity Management (high availability and disaster recovery) for a virtual environment
Securing a virtual environment (options and best practices)
When you install and configure virtualization for a SharePoint 2013 farm, every configuration choice that you make and implement has an effect on all other parts of the virtual environment. Your primary goal is to create intended consequences and positive effects. To achieve this goal you have to understand the interaction between the virtual and the physical, the inherent constraints. You also have to follow best practice guidance when you set up the virtual farm.
The first thing that you have to do before you install and configure your virtual environment is to make sure that you install the latest version of the Hyper-V Best Practices Analyzer (BPA). You can get this update from Update for Best Practices Analyzer for HYPER-V for Windows Server 2008 R2 x64 Edition (KB977238).
You can use Hyper-V Best Practices Analyzer to scan a server that is running the Hyper-V role and help identify configurations that do not comply with the best practices of Microsoft for this role. BPA scans the configuration of the physical computer, the virtual machines, and other resources, such as virtual networking and virtual storage. No configuration changes are made by running the scan.
The scan results display the following information:
A list of issues that you can sort by severity
Recommended fixes for issues
Links to instructions
The best practices in the following table explain several aspects of virtualization and are not necessarily specific to any virtualization technology.
Hyper-V partition terminology|
Hyper-V provides three types of partitions: the root partition, the parent partition, and the child partition. Each partition has its resources (memory and processor) and policies for device use. The root partition is the original partition and it starts the hypervisor. The parent partition calls the hypervisor to request the creation of new child partitions—the virtual machines.
Best practice guidance for virtualization
Leave adequate memory for the Hyper-V partitions.
Calculate the total memory requirements for all the virtual machines on the host and ensure that there is enough available memory to meet Hyper-V partition requirements.
Do not use the parent partition for services other than Hyper-V.
Do not run additional roles or services on the parent partition. Run them on the virtual machines instead because the parent is differentiated for scheduling.
Do not store host computer system files on drives that are used for Hyper-V storage.
To reduce disk contention, do not store system files (for example, Pagefile.sys) on hard disks that are dedicated to storing virtual machine data.
Use a minimum of two physical network adapters.
For better network management and performance, dedicate one adapter to virtual machine network traffic and use the other adapter for virtualization host network traffic.
Do not oversubscribe the CPU on the virtualization host computer.
Review the supported ratio of virtual processors per logical processor and avoid oversubscribing the host computer CPU. The optimum virtual processor:logical processor ratio is 1:1. For more information, see Configure the processors for the virtual machines.
Do not cross Non-uniform memory access (NUMA) boundaries.
Hyper-V spans NUMA nodes to assign physical memory to a virtual machine; however, this does reduce performance on the virtual machine. For more information, see Configure the memory for the virtual machines
Do not use snapshots in a production environment.
Do not use snapshots for the virtual machines in a SharePoint products production environment. When you create a snapshot, Hyper-V creates a new secondary drive for the virtual machine. Write operations occur on the new drive and read operations occur on both drives, which has the same net affect as a differencing disk. Every snapshot that you add reduces disk performance further.
When you configure the Hyper-V host computer, we recommend that you consider the following configuration options and best practices:
Close the operating system management windows.
Hyper-V Manager and Virtual Machine Connection sessions consume resources. For example, Hyper-V manager causes Windows Management Instrumentation (WMI) activity in the parent partition. When you close the Virtual Machine Connection Manager, video emulation is disabled, which eliminate this source of resource consumption.
Avoid running programs such as anti-virus software on the parent partition.
Run them on the child partition if they are required.
Use hardware that supports Second Level Address Translation (SLAT).
SLAT is hardware that is optimized for virtualization improves virtual machine performance, and reduces processing load on the Windows hypervisor. For more information, see Hyper-V: List of SLAT-Capable CPUs for Hosts.
If your hardware supports Hyper-Threading, enable it.
Hyper-Threading splits the CPU pipeline in two and makes a single core look like two cores.
Hyper-V provides the level of networking robustness and configuration options that you expect to see in a physical networking environment. There are, of course, some limitations because you are dealing with virtual devices.
We recommend that you refer to the following documentation to prepare to install and configure virtual networking.
You can create and configure a Hyper-V virtual network before you install and configure virtual machines. In addition, you can create more than one virtual network on a Hyper-V host computer.
|You cannot create more than one virtual network on a Hyper-V host computer if the computer is running a Server Core installation of Windows Server 2008 or Windows Server 2008 R2.|
Hyper-V provides three types of virtual networks that you can configure for virtual machines. The following table provides a summary of these virtual networks and their characteristics.
Virtual network types
Provides a communication link between virtual machines and a physical network by creating an association to a physical network adapter on the host computer. Dedicate one physical adapter to this type of network.
For security purposes you can isolate traffic between virtual machines and other computers on the physical network by clearing the Allow management operating system to share this network adapter setting. However, you will be unable to connect to the management operating system remotely.
Provides a communications link between the host computer and the virtual machines. This provides a degree of isolation from external network traffic and is typically used in a test environment where you want to connect to the virtual machines by using the management operating system.
Provides a communications link between the virtual machines. They are completely isolated from the host computer and this type of virtual network is often used to set isolated test domains.
|In a scenario where two internal (or private) virtual networks are created in Hyper-V and two virtual machines are created on a separate IP subnet, these virtual machines cannot communicate with one another. Because the virtual switch operates at layer 2 of the ISO/OSI Network Model, you have to have a router to achieve routing at a higher level.|
After you create your virtual network, you can specify the range of media access control (MAC) addresses that are automatically assigned to the virtual network adapters. Hyper-V enables you to provide static MAC addresses to a virtual adapter to avoid collisions on the network.
|John Howard's blog post, Hyper-V: MAC Address allocation and apparent network issues MAC collisions can cause provides a very good explanation about MAC Address allocation and associated network issues.|
Windows Server 2008 R2 adds the option to configure MAC address spoofing (Enable Spoofing Of MAC Addresses) in the virtual network adapter settings. For more information, see Configure MAC Address Spoofing for Virtual Network Adapters.
From a performance perspective, the ability to create virtual local area networks (VLANs) can provide significant throughput gains. Because virtual machines on the same VLAN can communicate through the virtual switch, network traffic is faster because it does not have to use the physical network adapter. Another benefit of a VLAN configuration is the fact that, because it is software-based, a virtual machine can easily be moved between hosts and still keeps its network configuration.
When you enable virtual LAN identification for the management operating system, you can assign a VLAN identifier (ID), which is an integer that uniquely identifies a node that belongs to a particular VLAN. If you use virtual LAN feature and a VLAN ID, note the following:
The physical adapters must support VLAN tagging and this feature has to be enabled.
Set the VLAN ID on either the virtual switch or on the virtual machine's network adapter instead of the physical adapter.
You can assign only one VLAN ID on the virtual switch.
Hyper-V provides two kinds of virtual network adapters that you can configure for a virtual machine: a network adapter and a legacy network adapter.
A network adapter, also known as a synthetic adapter, is the preferred option for most virtual machine configurations. The driver for this adapter is included with the integration services that are installed with the Windows Server 2008 R2 guest operating system.
A legacy adapter emulates an Intel 21140-based PCI Fast Ethernet Adapter, which results in a lower data transfer than the network adapter. A legacy network adapter also supports network-based installations because it can boot to the Pre-Boot Execution Environment (PXE).
Unless you have to use a legacy adapter until you can install the virtual machine driver or have to do a network boot, we recommend that you configure a virtual machine with a network adapter. If you do have to use a legacy adapter for a network installation, you can always add a network adapter later, and then delete the legacy adapter.
NIC teaming is the process of grouping several physical NICs into one logical NIC, which can be used for network fault tolerance and transmit load balance. The process of grouping NICs is called teaming. Teaming has two purposes:
Fault tolerance. Teaming more than one physical NIC to a logical NIC maximizes high availability. Even if one NIC fails, the network connection does not stop and continues to operate on other NICs.
Load balancing. Balancing the network traffic load on a server can enhance the functionality of the server and the network. Load balancing within network interconnect controller (NIC) teams enables distributing traffic among the members of a NIC team so that traffic is routed among all available paths.
|Windows Server 2008 Service Pack 2 (SP2) and Windows Server 2008 R2 have no restrictions that are associated with NIC Teaming and the Failover Clustering feature. In Windows Server 2008, the Microsoft Failover Cluster Virtual Adapter is compatible with NIC Teaming and enables it to be used on any network interface in a Failover Cluster.|
Windows Server 2008 R2 adds important new capabilities that you should use in your Hyper-V environment if your servers and network hardware support them. We recommend that you investigate the following networking options:
Large Send Offload (LSO) and Checksum Offload (CSO). The virtual networks in Hyper-V support LSO and CSO. In addition, if your physical network adapters support these capabilities, the virtual traffic is offloaded to the physical network as necessary. Most network adapters support LSO and CSO.
Jumbo frames. With Windows Server 2008 R2, jumbo frame improvements converge to support up to 6 times the payload per packet. This increases overall throughput and reduces CPU utilization for large file transfers. Physical networks and virtual networks support jumbo frames. This includes switches and adapters.
Note: For physical networks, all intervening network hardware such as switches must have jumbo frame support enabled also.
TCP chimney. This lets virtual NICs in child partitions to offload TCP connections to physical adapters that support it, which reduces CPU utilization and other overhead.
Virtual machine queue (VMQ). VMQ improves network throughput by distributing network traffic for multiple VMs across multiple processors. This process reduces processor utilization by offloading packet classification to the hardware and avoiding both network data copy and route lookup on transmit paths. VMQ is compatible with most other task offloads and can coexist with large send offload and jumbo frames.
Although Hyper-V supports several guest operating systems, SharePoint 2013 requires the 64-bit edition of Windows Server 2008 R2 Service Pack 1 (SP1) Standard, Enterprise, or Data Center. For more information about supported guest operating systems, see About Virtual Machines and Guest Operating Systems
Hyper-V provides many configuration options and you can change a configuration, the amount of memory for example, after the virtual machine is running as a SharePoint products farm server. With the exception of adding a virtual hard disk drive for a SCSI controller, you have to shut down a virtual machine before you can change its configuration.
Configure each virtual machine according to the requirements in Detailed design and system specification process for a virtual SharePoint 2013 farm. Configure the following for each virtual machine:
The BIOS setting to set the boot sequence (legacy network adapter, CD, IDE, or floppy disk)
The amount of memory
The number of virtual processors
The type and number of controllers
The type and number of hard disks
The type and number of network adapters
In addition to the previous configurations, you also have the option to configure a DVD drive, COM ports, and a virtual floppy disk.
From a SharePoint products perspective, the primary configuration considerations are the memory, processor, and hard disks.
Configure memory on a virtual machine as you typically do for an application that runs on a physical server. The memory allocation must be sufficient to reasonably handle the load at ordinary and peak times. For SharePoint virtual machines, insufficient memory is the main cause of performance issues.
Before you install and configure the virtual machines on a Hyper-V host computer, calculate how much memory is available for the virtual machines. The root partition must have sufficient memory to provide services such as I/O virtualization and management to support the child partitions. For SharePoint products, we recommend that you allow a minimum of 4 GB of RAM for overhead on a Hyper-V virtualization host computer.
After you factor in the 4 GB RAM reserve for the virtualization host, configure the virtual machines to use the remaining memory.
Windows Server 2008 R2 SP1 has the option of configuring dynamic memory (with a minimum value and maximum value) for virtual machines.
We do not support this option for virtual machines that run in a SharePoint 2013 environment. The reason is that this implementation of dynamic memory does not work with every SharePoint feature. For example, Distributed Cache and Search do not resize their caches when the allocated memory for a virtual machine is dynamically changed. This can cause performance degradation, especially when assigned memory is reduced.
Non-uniform memory access (NUMA)
A very import aspect of virtual machine memory configuration is Non-uniform memory access (NUMA). NUMA is a memory design that speeds up memory access by partitioning physical memory so each processor in a multi-CPU has its own memory. For example, in a system with 8 cores and 32 GB of RAM, each core or node has 4 GB of physical memory. If a virtual machine is configured to use 8 GB of RAM, the system has to use memory in another node. Because crossing the NUMA boundary can reduce virtual performance by as much as 8%, it is a best practice to configure a virtual machine to use resources from a single NUMA node. For more information about NUMA, refer to the following articles:
You can configure multiple virtual processors for a virtual machine, up to a limit of four processors. You cannot configure more processors per virtual machine than there are logical (cores) processors on the virtualization host. For example, given a dual core physical server, the limit is two virtual processors for a virtual machine. Although Hyper-V supports up to eight virtual processors per core or per thread, a configuration that exceeds this ratio (8:1) is known as being oversubscribed. For any virtual machine that you use in a SharePoint 2013 farm, we recommend a ratio of 1:1. Oversubscribing the CPU on the virtualization host can decrease performance, depending on how much the CPU is oversubscribed. For more information, see Hyper-V VM Density, VP:LP Ratio, Cores and Threads
You can use two controller options and several hard disk configurations for a virtual machine. Before you configure storage for your virtual machines, read the following posts written by Jose Barreto, who is a member of the File Server Team at Microsoft.
You can select either Integrated Device Electronics or SCSI devices on virtual machines, as follows:
IDE devices: Hyper-V uses emulated devices with IDE controllers. You can have up to two IDE controllers with two disks on each controller. The startup disk (also known as the boot disk) must be attached to one of the IDE devices. The startup disk can be either a virtual hard disk or a physical disk. Although a virtual machine must use an IDE device as the startup disk to start the guest operating system, you have many options to choose from when you select the physical device that will provide the storage for the IDE device.
SCSI devices: Each virtual machine supports up to 256 SCSI disks (four SCSI controllers with each controller supporting up to 64 disks). SCSI controllers use a type of device that was developed specifically for use with virtual machines and use the virtual machine bus to communicate. The virtual machine bus must be available when the guest operating system is started. Therefore, virtual hard disks that are attached to SCSI controllers cannot be used as startup disks.
Note: Physical SCSI devices typically provide better I/O performance than physical IDE devices. However, this is not the case for virtualized SCSI and IDE devices in Hyper-V. Support for hot swappable hard disk drives, which the Hyper-V implementation of SCSI supports, is a better reason for selecting SCSI drives than performance gains.
The version of Hyper-V released with Windows Server 2008 R2 provides significant improvements in virtual hard disk performance. For more information, see Virtual Hard Disk Performance: Windows Server 2008 / Windows Server 2008 R2 / Windows 7. For a summary of virtual machine drive options, see the "How to choose your Hyper-V and VHD Storage Container Format" section of " Virtual Hard Disk Performance: Windows Server 2008 / Windows Server 2008 R2 / Windows 7".
Hyper-V supports many storage options. For more information about the storage options, see Planning for Disks and Storage.
You can use the following types of physical storage with a server that runs Hyper-V:
Direct-attached storage: You can use Serial Advanced Technology Attachment (SATA), external Serial Advanced Technology Attachment (eSATA), Parallel Advanced Technology Attachment (PATA), Serial Attached SCSI (SAS), SCSI, USB, and Firewire.
Storage area networks (SANs): You can use Internet SCSI (iSCSI), Fibre Channel, and SAS technologies.
For more information, see Configuring Pass-through Disks in Hyper-V.
There is no generic storage solution for every virtual environment. Selecting the optimal virtual machine drive option for your SharePoint 2013 servers requires research and extensive testing to implement the best storage solution for your virtual environment. When you pick a storage solution you must consider access performance, storage needs, and how much memory is used for the advanced caching of virtual hard disk images.
In a Hyper-V environment, you can specify the configuration of virtual networking and the configuration for each virtual machine. Additionally, you can configure how each virtual machine interacts with the Hyper-V host computer, and also the stop and restart behavior if the running state of the virtual machine is interrupted.
Hyper-V includes a software package for supported guest operating systems that improves integration between the physical computer and the virtual machine. This package is known as integration services. You should verify that the management operating system (which runs the Hyper-V role) and virtual machines are running the same version of integration services. For more information, see Version Compatibility for Integration Services.
For each virtual machine you can configure the following integration items between the virtual machine and the virtualization host computer:
Operating system shutdown
Backup (volume snapshot)
|Disable the time synchronization for each SharePoint virtual machine. SharePoint 2013 implements timer jobs extensively and the latency during time synchronization will cause unpredictable results in the SharePoint environment.|
For each virtual machine you can configure automatic stop and start behavior if a physical computer shuts down.
The options for stop are as follows:
Save the virtual machine state.
The current state of the virtual machine is saved. When the virtual machine is started, Hyper-V attempts to restore the virtual machine to the state it was in.
Turn off the virtual machine.
This is the equivalent of pulling the power plug on a server.
Shut down the guest (virtual machine) operating system.
This is the equivalent of shutting down a computer by using the Windows Shut down option.
For a SharePoint products virtual machine, do not configure the virtual machine to save state. Virtual machines that start from saved state will be out of synchronization with the other servers in the farm. We recommend that you configure the virtual machine to use a shutdown because it minimizes that chances that the virtual machine can be corrupted. When a shutdown happens, all timer jobs that are running can finish, and there will be no synchronization issues when the virtual machine restarts.
The opposite of an automatic stop is an automatic start. Hyper-V provides the following startup options when the physical server restarts:
You have to start the virtual machine manually regardless of its state when the physical server shut down.
Automatically start if the virtual machine was running when the service stopped.
Always start this virtual machine automatically.
Hyper-V starts the virtual machine regardless of its state when the physical server shut down.
We recommend that you select either of the first two options. Both options are acceptable. However, the decision is ultimately up to the IT team that manages and maintains the virtual environment.
In addition to the previous start options, you can configure a startup time delay for a virtual machine. We recommend that you do this to reduce resource contention on a virtualization host. However, if your start option is to do nothing, this is not an issue.